To investigate potential differences between childhood-onset and adult-onset systemic lupus erythematosus (SLE).
To investigate potential differences between childhood-onset and adult-onset systemic lupus erythematosus (SLE).
An inception cohort with childhood-onset SLE (n = 67) was compared with an inception cohort with adult-onset SLE (n = 131), each of whom was diagnosed between 1990 and 1998 and followed up until February 1999. Prospective information included data on medications, laboratory markers, and disease activity and damage as measured by the SLE Disease Activity Index (SLEDAI) and the Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index (SDI), respectively.
Eighty-five percent of patients with childhood-onset SLE and 88% of patients with adult-onset SLE were female; the mean duration of followup was 3.2 and 3.5 years, respectively. On average, the children had more-active disease than did the adults at the time of diagnosis and during followup. There was a higher incidence of renal disease in those with childhood-onset SLE (78% versus 52% in adults; P = 0.0005), and the adjusted mean SLEDAI renal score was higher in the children than in the adults (2.37 versus 0.82; P < 0.0001). Treatment with steroids (97% versus 72%; P < 0.0001) and immunosuppressive drugs (66% versus 37%; P = 0.0001) was used significantly more often in children with SLE. Four adult patients with SLE, but none of the children, died during the followup. At the end of the followup, the mean SDI scores in those with childhood-onset SLE were higher than those with adult-onset SLE (1.70 versus 0.76; P = 0.008).
Children with childhood-onset SLE have more active disease at presentation and over time than do adults with SLE, especially active renal disease. Compared with adults with SLE, children receive more intensive drug therapy and accrue more damage, often related to steroid toxicity.
Of all patients with systemic lupus erythematosus (SLE), 15–20% are diagnosed during childhood (1), with disease onset prior to the age of 16 years. Previous studies using nonstandardized disease measures suggest that immunologic, serologic, and clinical abnormalities of children with SLE are more pronounced than those of adults. In recent years, disease indices have been developed that allow for a standardized comparison of pediatric and adult SLE cohorts (2–5). Using these standardized disease measures, reports from several large prospective cohorts of adults and some pediatric cohorts suggest that SLE in adults is less active and is associated with less permanent damage than is childhood-onset SLE (6–14).
Because disease expression in SLE is influenced by environmental factors and differs between racial and ethnic groups, it is important to use cohorts of adults and children with SLE from similar backgrounds (6, 7, 15–19). The objectives of our study were as follows: to compare adult-onset SLE patients and childhood-onset SLE patients who received followup care in the same academic center and had similar social and environmental backgrounds for differences in overall and renal disease activity, and to compare the 2 groups for differences in the type and amount of disease damage according to standardized disease indices.
An inception cohort of all childhood-onset SLE patients at the Lupus Clinic of the Hospital for Sick Children in Toronto was compared with an inception cohort of adult-onset SLE patients at the University of Toronto Lupus Clinic, University Health Network, Toronto Western Hospital. All patients were entered into their respective cohorts within 1 year of diagnosis. Sixty-seven children and 132 adults with SLE who were entered into the cohorts at the 2 clinics between January 1990 and December 1998 were studied. All pediatric SLE patients were followed up at the Hospital for Sick Children in Toronto. No patient receiving followup care in other Canadian provinces was included, and all patients lived within driving distance of the hospital. The childhood-onset SLE patients were examined at 2-month intervals and the adults at 4-month intervals.
Patients of both groups fulfilled at least 4 of the American College of Rheumatology (ACR) classification criteria for SLE (20). All pediatric and adult SLE patients have been followed up according to a prospective protocol. Findings and data obtained until February 1999 were considered in the study. Clinical information was recorded on standardized data collection forms. Information regarding medication use and the results of the initial renal biopsy analyses were recorded. A standardized prospective laboratory panel, including anticardiolipin antibody testing, was obtained in patients at both centers.
Based on the standardized data collected, we determined the overall and renal disease activity, as well as disease damage, in all patients. Disease activity was measured by the SLE Disease Activity Index (SLEDAI). Disease damage was measured by the Systemic Lupus International Collaborating Clinics/ACR Damage Index (SDI).
SLEDAI assessment. The SLEDAI is a commonly used disease activity measure for adult-onset SLE and childhood-onset SLE (2, 21–28). The tool consists of 24 weighted items grouped into 9 domains, or organ systems, as follows: central nervous system (assigned a weight of 8), vascular (weight of 8), renal (weight of 4), musculoskeletal (weight of 4), serosal (weight of 2), dermal (weight of 2), immunologic (weight of 2), constitutional (weight of 1), and hematologic (weight of 1). If during the 10-day period prior to the assessment, a patient fulfills an attribute, then the corresponding weighted score will be given. The sum of the weighted domain scores comprises the overall SLEDAI disease activity score. SLEDAI scores range between 0 and 105, with 0 being inactive disease.
Measurement of renal disease activity was based on the weighted items of the renal domain only. The presence of clinical renal disease was defined as a SLEDAI renal domain score >0 during the followup period.
SDI assessment. The SDI is the only available disease damage index for SLE that has been widely validated for use in children and adults with SLE (3, 23, 24, 29–31). The SDI measures irreversible organ damage since the time of diagnosis of SLE. Damage scored in the SDI is grouped into 12 different organ systems, as follows: ocular (maximum score 2), neuropsychiatric (maximum score 6), renal (maximum score 3), pulmonary (maximum score 5), cardiovascular (maximum score 6), peripheral vascular (maximum score 5), gastrointestinal (maximum score 6), musculoskeletal (maximum score 7), and skin (maximum score 3). Damage scores are also given for the presence of premature gonadal failure (maximum score 1), diabetes mellitus (maximum score 1), and malignancy (maximum score 2). To be considered in the SDI, most items must persist for at least 6 months. Items such as cataracts, avascular necrosis, or myocardial infarction are counted when they occur. An SDI score of 0 is given for a patient who demonstrates none of the SDI items. The maximum SDI score is 47. Steroid-related damage items considered in the analysis were cataracts, osteoporotic fractures, and avascular necrosis.
SAS 9.0 software was used for the statistical analyses (SAS Institute, Cary, NC). The burden of ongoing disease activity under consideration of the time was determined by the adjusted mean SLEDAI (AMS) score (32). The AMS score is defined by the area under the curve (33) of the SLEDAI scores over time by adding the area of each of the blocks of visit interval and then dividing them by the length of time for the whole period. Similarly, for measuring the burden of active renal disease activity over time, the AMSrenal score was calculated using the algorithm described above and the SLEDAI scores from the renal domain only. The AMS score is particularly suitable for this study since patients may have attended their respective clinics at irregular intervals, while the total duration of followup was similar in the adult and pediatric groups (32). The cohorts were compared using 2-sample t-tests, chi-square tests, and Fisher's exact tests. A linear regression model was used to assess the difference between childhood-onset and adult-onset SLE in terms of disease activity and damage after correcting for differences in the proportions of Caucasian/non-Caucasian race and male/female sex in each group, as well as the use of steroids and immunosuppressive agents. To correct for multiple comparisons between cohorts, only Bonferroni-corrected P values less than or equal to 0.004 were deemed statistically significant.
Sixty-seven children and 131 adults with SLE who were diagnosed between 1990 and 1998 were studied. Disease followup information until February 1999 was included. The cohort of childhood-onset SLE patients consisted of 57 female and 10 male patients with a mean ± SD age at the time of diagnosis of 12.7 ± 2.5 years (range 3.8–16.8 years). The cohort of adult-onset SLE patients consisted of 116 female and 15 male patients, with a mean ± SD age at the time of diagnosis of 36.0 ± 13.2 years (17.1–74.8 years). The mean ± SD disease duration at the first clinic visit was 1.13 ± 5.01 months (range 0–36.0 months) in the childhood-onset SLE group and 2.83 ± 3.43 months (range 0–12.9 months) in the adult-onset SLE group (P = 0.014). The duration of followup was similar in the two cohorts, with a mean ± SD of 3.2 ± 2.0 years (range 0.3–7.9 years) in the childhood-onset SLE group and 3.5 ± 2.6 years (range 0–8.7 years) in the adult-onset SLE group (P not significant).
Since the majority of patients in both cohorts were Caucasian, we grouped the patients of other ethnicities into a non-Caucasian group. Although there were more Caucasian patients in the adult-onset SLE cohort than the childhood-onset SLE cohort, this difference did not reach statistical significance (Caucasian:non-Caucasian patients 96:35 with adult-onset SLE and 41:26 with childhood-onset SLE; P not significant). Elevated levels of anticardiolipin antibodies were present in 32 of the 44 children tested (73%) and in 76 of the 124 adults tested (61%) (P not significant). Unfortunately, data on other autoantibodies was not electronically recorded for the pediatric group; thus, it was not possible to analyze other autoantibody levels in the two groups.
The decision to perform a kidney biopsy was based on the presence of clinical signs consistent with renal disease. Forty-three of the childhood-onset SLE patients (64%), but only 24 of the adults (18%), underwent kidney biopsy at least once in the first 3 years of followup. There were no significant differences in the findings of the renal biopsies between groups, although there was a trend for more patients in the pediatric cohort to have severe renal involvement (World Health Organization class III and IV ), with 66% of childhood-onset SLE patients having these findings on renal biopsy as compared with 50% of the adults (P not significant) (Table 1).
|Childhood-onset SLE (n = 67)||Adult-onset SLE (n = 131)|
|Patients with any renal involvement†||52/67 (78)||68/131 (52)‡|
|Patients with at least 1 renal biopsy||43/67 (64)||24/131 (18)|
|WHO classification of the first renal biopsy|
|Class II||10/43 (23)||5/22 (23)|
|Class III||11/43 (26)||4/22 (18)|
|Class IV||17/43 (40)||7/22 (32)|
|Class V||5/43 (11)||6/22 (27)|
The vast majority of patients with childhood-onset SLE received oral corticosteroids for disease control, whereas this medication was used significantly less frequently in the adults (97% versus 70%; P < 0.0001) (Table 2). Similarly, childhood-onset SLE patients were significantly more often treated with high-dose intravenous methylprednisolone than were the adults (30% versus 11%; P < 0.001) (Table 2). There was no difference in the frequency of treatment with antimalarial medications. Immunosuppressive medications were used in 66% of those with childhood-onset SLE, but only 37% of those with adult-onset SLE (P = 0.0001). Only the use of methotrexate differed between the 2 groups (31% in adults and 9% in children; P = 0.009)
|Childhood-onset SLE (n = 67)||Adult-onset SLE (n = 131)||P|
|Oral corticosteroids||65 (97)||92 (70)||<0.0001|
|Pulse methylprednisolone||20 (30)||15 (11)||0.001|
|Antimalarial agents||54 (81)||95 (73)||0.21|
|Immunosuppressive medications||44 (66)||48 (37)||0.0001|
|Azathioprine||28 (64)||36 (75)||0.24|
|Cyclophosphamide||11 (25)||10 (21)||0.63|
|Methotrexate||4 (9)||15 (31)||0.009|
|Mycophenolate mofetil||0||1 (2)||1.00|
Children with SLE had higher mean SLEDAI scores at presentation than did the adults with SLE (16.8 ± 10.1 versus 9.3 ± 7.6; P < 0.0001) (Table 3). Adults with SLE achieved better disease control than did children with SLE during the 3 years of followup, as supported by significantly lower mean AMS scores (Table 3). Children with SLE continued to have significantly more active renal disease over time (AMSrenal score 2.37 versus 0.82; P < 0.0001) (Table 3). In exploratory analyses using a linear regression model to correct for sex, Caucasian race, steroid use, and immunosuppressive agent use between groups, patients with childhood-onset SLE still had significantly more activity at presentation and active renal disease during followup than did those with adult-onset SLE (P < 0.0001). The other risk factor associated with increased activity was the use of immunosuppressive agents (P = 0.01 for the SLEDAI score at presentation; P < 0.0001 for the AMS score and for the AMSrenal score).
|Childhood-onset SLE (n = 67)||Adult-onset SLE (n = 131)||P|
|Maximum SLEDAI score in cohort||69||45|
|SLEDAI score at diagnosis||16.8 ± 10.1 (0–44)||9.3 ± 7.6 (0–45)||<0.0001|
|Adjusted mean SLEDAI score||5.70 ± 2.68 (1.3–13.5)||4.59 ± 3.05 (0–18.2)||0.012|
|Adjusted mean renal SLEDAI score||2.37 ± 2.38 (0–9.7)||0.82 ± 1.59 (0–8.7)||<0.0001|
None of the patients with childhood-onset SLE died during followup, but 4 of the patients with adult-onset SLE died (P = 0.3). The causes of death in these 4 patients were active lupus, gastrointestinal hemorrhage, renal failure and lymphoma, and infection.
Minimal disease duration of 6 months is necessary for the initial scoring of the SDI. Thirty-seven (56.1%) of the pediatric patients and 57 (43.5%) of the adult patients had an SDI score > 0, representing any disease damage, at the end of the followup period (P not significant) (Table 4). Based on the mean SDI scores, patients with childhood-onset SLE had a greater amount of damage than did patients with adult-onset SLE (1.70 versus 0.76; P = 0.008).
|Childhood-onset SLE (n = 66)||Adult-onset SLE (n = 131)||P|
|SDI score, mean ± SD (range)||0.48 ± 0.61 (0–2)||0.14 ± 0.37 (0–2)||<0.0001†|
|No. (%) with SDI score >0||28 (42.4)||17 (13.0)||<0.0001|
|SDI score, mean ± SD (range)||0.30 ± 0.96 (0–5)||0.11 ± 0.33 (0–2)||0.11|
|No. (%) with SDI score >0||8 (12.1)||13 (9.9)||0.64|
|SDI score, mean ± SD (range)||0.17 ± 0.60 (0–3)||0.08 ± 0.34 (0–3)||0.26|
|No. (%) with SDI score >0||6 (9.1)||8 (6.1)||0.56|
|SDI score, mean ± SD (range)||0.03 ± 0.17 (0–1)||0.02 ± 0.15 (0–1)||0.76|
|No. (%) with SDI score >0||2 (3.0)||3 (2.3)||1.00|
|SDI score, mean ± SD (range)||0.02 ± 0.12 (0–1)||0.05 ± 0.26 (0–2)||0.16|
|No. (%) with SDI score >0||1 (1.5)||6 (4.6)||0.43|
|Peripheral vascular damage|
|SDI score, mean ± SD (range)||0.05 ± 0.27 (0–2)||0.02 ± 0.19 (0–2)||0.55|
|No. (%) with SDI score >0||2 (3.0)||2 (1.5)||0.60|
|SDI score, mean ± SD (range)||0.06 ± 0.35 (0–2)||0.02 ± 0.15 (0–1)||0.40|
|No. (%) with SDI score >0||2 (3.0)||3 (2.3)||1.00|
|SDI score, mean ± SD (range)||0.48 ± 0.93 (0–4)||0.13 ± 0.42 (0–2)||0.004|
|No. (%) with SDI score >0||16 (24.2)||13 (9.9)||0.007|
|SDI score, mean ± SD (range)||0.08 ± 0.27 (0–1)||0.08 ± 0.33 (0–2)||0.86|
|No. (%) with SDI score >0||5 (7.6)||9 (6.9)||1.00|
|Premature gonadal failure damage|
|SDI score, mean ± SD (range)||0 ± 0||0.02 ± 0.12 (0–1)||0.16|
|No. (%) with SDI score >0||0 (0)||2 (1.5)||0.55|
|SDI score, mean ± SD (range)||0.03 ± 0.17 (0–1)||0.05 ± 0.21 (0–1)||0.61|
|No. (%) with SDI score >0||2 (3.0)||6 (4.6)||0.72|
|SDI score, mean ± SD (range)||0 ± 0||0.04 ± 0.19 (0–1)||0.02|
|No. (%) with SDI score >0||0 (0)||5 (3.8)||0.17|
|Mean SDI score at the end of followup|
|SDI score, mean ± SD (range)||1.70 ± 2.67 (0–12)||0.76 ± 1.16 (0–7)||0.008|
|No. (%) with SDI score >0||37 (56.1)||57 (43.5)||0.10|
Compared with the adults, the children had significantly more steroid-associated damage, including cataracts (42% versus 12%; P < 0.0001) and avascular necrosis (21% versus 5%; P = 0.0003), but not osteoporotic fractures (6% versus 3%; P not significant). In the 37 patients with childhood-onset SLE and the 57 patients with SLE with damage, 50.9% of the damage observed in childhood-onset SLE patients and 29.3% of the damage observed in adult-onset SLE appeared to be steroid-related (P = 0.001). In a linear regression model adjusting for sex, Caucasian race, and use of steroids and of immunosuppressive agents, childhood-onset SLE was associated with increased damage (P = 0.04) as well as with the use of immunosuppressive agents (P < 0.0001).
The results of our study support the concept that SLE in children is more active and is associated with more rapid accrual of damage than is SLE in adults. In our study, the racial distribution in the pediatric group as compared with that in the adult group was similar, unlike in previous studies (14). Our patients were drawn from the same geographic area, using prospective longitudinal data with standardized disease measures, which decreases the reliability and validity of the comparisons made. A comparable racial distribution of childhood-onset SLE and adult-onset SLE patients living in the same geographic area has previously been reported from Brazil (12). Similar to the findings of previous studies, our study was inconclusive as to whether anticardiolipin antibody positivity is more common in childhood-onset SLE than in adult-onset SLE (12–14).
A higher frequency of renal disease and a concomitantly lower frequency of cardiopulmonary involvement have been reported in other studies that compared childhood-onset SLE and adult-onset SLE (12, 14). In our study, a higher percentage of pediatric patients had biopsy-proven renal disease, and there was a trend toward significantly more proliferative nephritis in this group. This increased frequency and severity of renal disease was associated with increased disease activity, which was primarily observed in the renal domain rather than in other organ systems, in pediatric patients, confirming previous observations that renal disease is more common and more severe in childhood-onset SLE than in adult-onset SLE. This also supports the findings of an earlier study from Thailand, which compared 51 children and 308 adults with SLE and showed that there was a higher percentage of patients with renal disease in the childhood-onset SLE group as compared with the adult-onset SLE group (80% versus 53%) (13). Despite the more frequent use of immunosuppressive agents, the renal AMS score was more than twice as high among the children as the adults with SLE. In addition, global disease activity was higher in patients with childhood-onset SLE as manifested by a higher AMS score. Significant differences in the activity of renal disease (AMSrenal score) and overall disease activity (AMS score) between adult-onset SLE and childhood-onset SLE were also noted among patients of similar racial groups (Caucasian and non-Caucasian). These results suggest that current medications and therapeutic regimens tested mostly in adults may not be sufficient to achieve comparable control of overall disease activity and renal disease activity in patients with childhood-onset SLE.
Our results suggest that, compared with adult SLE patients, children with childhood-onset SLE had more active disease at diagnosis and developed damage over time more rapidly. These findings support the notion that high levels of disease activity at diagnosis, whether in childhood or adult SLE, are a risk factor for poor outcome (11, 32, 35).
Although a number of items in the SDI may reflect aging (cataracts, myocardial infarction, diabetes mellitus, malignancy), patients with childhood-onset SLE still developed significantly more disease damage than did the adult patients. This is likely related to increased disease activity and higher levels of corticosteroids and immunosuppressive agents required for disease control in the children as compared with the adults. Of note, most of the difference in the accrual of damage between children and adults with SLE was due to steroid-related damage features. In other pediatric cohorts, such as the one studied by Ravelli et al (36), both the overall damage and the steroid-related damage were less than that in our childhood-onset SLE patients. However, that study was a retrospective analysis, and a higher percentage of the patients were Caucasians, who are known to have less active disease and to accrue less disease damage over time than non-Caucasian patients (9, 37). The mortality rate in our cohort of children was lower than that reported in earlier childhood-onset SLE cohorts and was lower than that seen in our adult cohort or in other reports of adults with SLE (12, 14, 38).
Reasons for greater disease activity and more damage in childhood-onset SLE remain to be elucidated. Hormonal changes observed during puberty might add to the imbalance of the immune system in children with SLE. We did not see any difference in damage scores between prepubertal and postpubertal children (data not shown); therefore, we cannot address this issue in our study.
Limitations of our study include the fact that it is unlikely all adult SLE patients who were diagnosed in the Greater Toronto area during 1990 through 1998 were captured in the study. Because the Lupus Clinic at the Toronto Western Hospital is a well-known primary, secondary, and tertiary center for patients with SLE, we likely missed some of the less severely affected adult SLE patients. Based on the Canadian health system, we are very confident that almost all pediatric patients were included in the study. Thus, the finding of the study that childhood-onset SLE is more aggressive than adult-onset SLE would probably have been even more pronounced if all adults with SLE had been included.
Another limitation is that premature gonadal failure might have been missed in some of the childhood-onset SLE patients. In the SDI, premature gonadal failure is defined by the presence of secondary amenorrhea for at least 6 months. However, many of the childhood-onset SLE patients had not yet had regular menses, making the reliable assessment of this SDI item difficult. However, none of the postpubertal female adolescents had any evidence of secondary amenorrhea.
Study information on all of the adult SLE patients was recorded in a prospective database, whereas study information on the childhood-onset SLE patients was extracted retrospectively from the medical charts. However, all pediatric patient visit information was based on the use of standardized clinic forms, which had been developed to accurately record all of the information required to determine both the SLEDAI and the SDI scores. In addition, all childhood-onset and adult-onset SLE patients had regular standard laboratory testing. Therefore, our data are equivalent to high-quality prospectively collected data.
In summary, this is the first prospective study comparing adults and children with SLE from the same ethnic and environmental backgrounds using standardized disease measures, allowing for a valid comparison of adult-onset SLE with childhood-onset SLE. Compared with adult SLE patients, children with SLE have more active disease at diagnosis and over time. They acquire renal disease at a higher frequency and develop damage more rapidly, but the mortality rates in our cohorts were similarly low. The exact mechanisms of the more aggressive SLE in children than in adults remain to be elucidated.
Dr. Brunner, Dr. Gladman, and Ms Ibañez had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. Brunner, Gladman, Urowitz, Silverman.
Acquisition of data. Brunner, Gladman, Urowitz, Silverman.
Analysis and interpretation of data. Brunner, Gladman, Ibañez, Urowitz, Silverman.
Manuscript preparation. Brunner, Gladman, Urowitz, Silverman.
Statistical analysis. Ibañez.